基于支持向量机的车辆驾驶行为识别研究

发布时间：2018-05-09 21:46

本文选题：交通工程 + 驾驶行为识别　；参考：《交通运输系统工程与信息》2017年01期

【摘要】：从车辆行驶轨迹的角度,车辆驾驶行为可细分为车辆跟驰行为、车辆换道准备行为和车辆换道执行行为,它们对交通拥堵、交通事故等都有着重要影响,也是自动驾驶、交通仿真等系统的基础构成模块.然而,如何从实际微观交通流数据中对3种行为进行识别是驾驶行为研究的基础和难点.本文提出基于支持向量机的驾驶行为识别方法,使用真实车辆轨迹数据,为提高模型的准确率,首先对样本数据进行归一化和主成分分析预处理,然后采用网格搜索算法对惩罚因子和核参数进行寻优,最后利用样本数据对基于支持向量机的分类模型进行训练和测试.结果表明,模型的测试精度达到了98.41%,能够很好地识别车辆的行驶状态,为驾驶行为各阶段的研究提供支持.
[Abstract]:From the point of view of vehicle trajectory, vehicle driving behavior can be subdivided into car-following behavior, vehicle-changing preparation behavior and vehicle-changing execution behavior, which have an important impact on traffic congestion, traffic accidents and so on, and are also self-driving. Traffic simulation and other basic components of the system. However, how to identify the three behaviors from the actual microscopic traffic flow data is the foundation and difficulty of driving behavior research. In this paper, a driving behavior recognition method based on support vector machine (SVM) is proposed. In order to improve the accuracy of the model, the sample data are normalized and preprocessed by principal component analysis (PCA) in order to improve the accuracy of the model by using the real vehicle trajectory data. Then the mesh search algorithm is used to optimize the penalty factor and kernel parameters. Finally, the classification model based on support vector machine is trained and tested using the sample data. The results show that the accuracy of the model is 98.41, which can identify the driving state of the vehicle and provide support for the study of driving behavior.
【作者单位】：西南交通大学交通运输与物流学院;
【基金】：国家自然科学基金(51408509) 中央高校基本业务经费(2682016CX046)~~
【分类号】：U491.25

【相似文献】